On the (fast) quenching of (young) low-mass galaxies to
New spotlights on environment lead role
We investigate the connection between environment and the different quenching channels that galaxies are prone to follow in the rest-frame NUVrK (i.e., NUV-r vs. r-K) colour diagram, as identified by Moutard et al. (2016b). Namely, the (fast) quenching channel followed by (young) low-mass galaxies and the (slow) quenching channel followed by (old) high-mass ones. We make use of the >22 deg covered the VIPERS Multi-Lambda Survey (VIPERS-MLS) to select a galaxy sample complete down to stellar masses of at ( at ) and including 33,500 (43,000) quiescent galaxies properly selected at , while being characterized by reliable photometric redshifts () that we use to measure galaxy local densities. We find that (1) the quiescence of low-mass  galaxies requires a strong increase of the local density, which confirms the lead role played by environment in their fast quenching and, therefore, confirms that the low-mass upturn observed in the stellar mass function of quiescent galaxies is due to environmental quenching. We also observe that (2) the reservoir of low-mass galaxies prone to environmental quenching has grown between and whilst the share of low-mass galaxies in the quiescent population may have simultaneously increased, which may be consistent with a rising importance of environmental quenching with cosmic time, compared to mass quenching. We finally discuss the composite picture of such environmental quenching of low-mass galaxies and, in particular, how this picture may be consistent with a delayed-then-rapid quenching scenario.
keywords:galaxies: photometry – galaxies: distances and redshifts – galaxies: statistics – galaxies: interactions – galaxies: star formation – galaxies: evolution
The fact that galaxies can be classified according to their star-formation activity into a blue/star-forming population, mostly made of disc galaxies, and a red/quiescent population, mainly consisting of elliptical galaxies, has been extensively documented in the last decade (e.g., Hogg et al., 2003; Kauffmann et al., 2004; Baldry et al., 2006; Haines et al., 2007; Williams et al., 2009; Arnouts et al., 2013; Moutard et al., 2016a; Pacifici et al., 2016a). This bimodality, which can be observed to redshift (e.g., Ilbert et al., 2013; Muzzin et al., 2013; Tomczak et al., 2014; Mortlock et al., 2015; Davidzon et al., 2017), is the statistical expression of a fairly rapid phenomenon of star-formation shutdown, the so-called quenching. The processes that are involved in such quenching of star formation are, however, still a matter of debate. In particular, the quenching mechanism(s) that turn(s) star formation off in low-mass galaxies may be quite different from what is at play in massive galaxies.
Now well established, the predominance of the quiescence in massive galaxies (see, e.g., Bundy et al., 2006; Ilbert et al., 2010; Baldry et al., 2012; Davidzon et al., 2013; Moutard et al., 2016b) underlies a downsizing of the star-formation quenching (i.e., the more massive a galaxy is, the earlier its star formation stops, on average). Furthermore, the high constancy of the stellar mass function (SMF) of star-forming galaxies at high mass supports the idea that star-formation activity is preferentially impeded above a given stellar mass (i.e., the star-formation efficiency declines exponentially above this stellar mass; Ilbert et al., 2010; Peng et al., 2010), which has been confirmed to be remarkably stable over several Gyrs from (namely, at ; Moutard et al., 2016b). Actually, this characteristic stellar mass may also be considered as a dark matter halo critical mass of (assuming a stellar-to-halo mass ratio; e.g., Coupon et al., 2015). This may be consistent with virial shock-heating processes (e.g., Kereš et al., 2005; Dekel & Birnboim, 2006; Cattaneo et al., 2006), but other mechanisms able to halt the cold-gas supply such as feedback from a radio-loud active galactic nucleus (AGN) may also explain the star-formation quenching in massive galaxies (e.g., Best et al., 2005; Croton et al., 2006; Karouzos et al., 2014), which appears to be characterised by quite long timescales (of 1-to-a few Gyrs) over the last ten Gyrs (e.g., Schawinski et al., 2014; Ilbert et al., 2015; Moutard et al., 2016b; Pandya et al., 2017). However, such mass quenching processes can not be invoked in low-mass galaxies, and environmental effects have been put forth to explain the star-formation suppression in these galaxies.
Indeed, the latest measurements of the SMF reveal a clear excess of low-mass quiescent galaxies, which underlies an upturn around stellar masses of observed in the local Universe (e.g., Baldry et al., 2012; Moustakas et al., 2013) and at low redshift (Drory et al., 2009; Moutard et al., 2016b, to ), and whose build-up is observed at higher redshift (e.g., Muzzin et al., 2013; Tomczak et al., 2014, to for the later). We recently showed that quiescent galaxies that are responsible for this low-mass upturn in the SMF are young quiescent galaxies (Moutard et al., 2016b) –i.e. they exhibit colours typical of young stellar populations (making them good candidates to being post-starburst galaxies; e.g., Kriek et al., 2010; Whitaker et al., 2012)– and are expected to have experienced a rapid quenching (turning quiescent over just Gyr). Such observations support a picture where galaxies follow different quenching channels depending on their stellar mass, which is consistent with a scenario mixing different modes of star-formation quenching as proposed by Faber et al. (2007). In particular, the excess of low-mass quiescent galaxies has been suggested to be associated with the environmental quenching of satellite galaxies, whose importance is expected to grow with large-scale structure and, thus, to decrease with redshift (Peng et al., 2010).
The connection between environment and star-formation quenching is now well illustrated at low redshift (; e.g., Balogh et al., 1997; Lewis et al., 2002; Hogg et al., 2003; Kauffmann et al., 2004; Baldry et al., 2006; Haines et al., 2007; Yang et al., 2009; Peng et al., 2012), and a clear picture has emerged where, on average, red/quiescent galaxies are characterised by richer/denser environments than blue/star-forming ones. Several quenching processes involving rich environments have therefore been proposed, such as ram-pressure stripping, in which the gas is expelled from the galaxy that becomes satellite (Gunn & Gott, 1972); strangulation/starvation,111We emphasise that the terms strangulation/starvation might either refer to environment (e.g., when a galaxy enters the hot gas of a cluster) or to peculiar evolution (e.g., when the radio-loud AGN feedback halts the cold gas infall). in which the cold gas supply is heated and then halted (Larson et al., 1980; Peng et al., 2015); galaxy harassment, in which multiple encounters deprive galaxies from stars and/or gas through tidal stripping (Farouki & Shapiro, 1981; Moore et al., 1996); or major merging triggering a subsequent starburst episode and/or an AGN that consumes/expels the remaining reservoir of cold gas (Schawinski et al., 2014); all assuming that cold gas fuelling is impeded in dense environments. This stresses indeed that these processes must be addressed in the cosmological context of the hierarchical growth of large-scale structures, and especially the evolution of filaments along which flows the cold gas that fuels star formation inside galaxies (Sancisi et al., 2008; Dekel et al., 2009).
Much effort has been made over the last decade to observe the impact of environment on star formation across cosmic time (e.g., Cucciati et al., 2006, 2010; Muzzin et al., 2012; Lani et al., 2013; Scoville et al., 2013; Muzzin et al., 2014; Fossati et al., 2017; Cucciati et al., 2017; Malavasi et al., 2017; Laigle et al., 2018). However, for different reasons, these studies focused on relatively massive galaxies and were not able to probe a low-mass population whose prime interest is precisely the fact that, by "nature", it is not expected to quench. While the impact of environment on the quenching of low-mass quiescent galaxies has been observed for a while in the local Universe (Hogg et al., 2003) where these galaxies have appeared to be essentially satellites (e.g., Haines et al., 2007; Peng et al., 2012), the impact of environment on the quenching of low-mass galaxies was not observed at higher redshift until recently (namely, at ; Guo et al., 2017). This reemphasized the question of the impact of environment on the quenching of low-mass galaxies across cosmic time while raising the question of the associated contribution to the build-up of the quiescent population, in particular, in the light of the picture described previously where galaxies are prone to follow different quenching channels depending on their stellar mass.
In this paper, we analysed the relation between quenching and environment aiming, in particular, to verify whether environment drives the fast quenching channel followed by low-mass galaxies and responsible for the upturn observed in the SMF of quiescent galaxies, as shown in Moutard et al. (2016b). At the same time, we took this opportunity to question the importance of such quenching channel across cosmic time, compared to the quenching channel that can be associated with mass quenching. We made use of the unique combination of area, depth and photometric multi-wavelength coverage of the VIPERS Multi-Lambda Survey222http://cesam.lam.fr/vipers-mls/ (VIPERS-MLS; Moutard et al., 2016a), assembled in the fields of the VIMOS Public Extragalactic Redshift Survey333http://vipers.inaf.it/ (VIPERS; Guzzo et al., 2014). Covering >22 deg down to , the VIPERS-MLS is indeed remarkable as (a) it allows the use of the rest-frame NUV-r vs. r-K (NUVrK) diagram to properly separate quiescent and star-forming galaxies; (b) it provides a complete sample of galaxies down to stellar masses of at ( at ) including more than 33,500 (43,000) quiescent galaxies, which enabled us to probe the evolution of fairly low-mass galaxies from ; while (c) these galaxies are all characterised by accurate photometric redshifts, with , which allows for reliable local density measurements.
The paper is organised as follows. In Sect. 2 we give an overview of the VIPERS-MLS data and measurements used in the present study. We then review the NUVrK diagram and its ability to distinguish between fast and slow quenching channels in Sect. 3. In Sect. 4 we present our results regarding the connection between environment and quenching channels to finally discuss these results in Sect. 5.
2 Data: VIPERS-MLS
Observational data, photometric redshifts and stellar mass estimates were discussed extensively in Moutard et al. (2016a, b) and here we only present a brief overview of key elements (Sect. 2.1). To these preexisting measurements, we have now also added the measurement of local galaxy density, as described in Sect. 2.2.
2.1 Observational data, photometric redshifts, and mass estimates
Our data consist of (FUV, NUV,) u, g, r, i, z and imaging of 22.38 deg –after masking and quality cuts– within the VIPERS-MLS (Moutard et al., 2016a), a follow-up program in the fields of the spectroscopic survey VIPERS (Guzzo et al., 2014), i.e., in the fields W1 and W4 of the Canada-France-Hawaii Telescope Legacy Survey444http://www.cfht.hawaii.edu/Science/CFHTLS/ (CFHTLS). The VIPERS-MLS optical imaging has been based on the CFHTLS T0007 release (Hudelot et al., 2012) that reaches 80% completeness depth to i, while the -band data were obtained through new observations reaching over deg. The multi-wavelength coverage of VIPERS-MLS has been complemented by GALEX (Martin & GALEX Team, 2005) FUV and NUV data combining preexisting and new observations over deg, incorporated after using u-band images as priors. For full details of the data processing and catalogue creation see Moutard et al. (2016a).
Photometric redshifts were derived as described in Moutard et al. (2016a) by using the template-fitting code Le Phare (Arnouts et al., 2002; Ilbert et al., 2006). Photometric redshift (photo-z) estimates were validated using extensive VIPERS spectroscopy (90,000 spectroscopic redshifts to i ; Scodeggio et al., 2018), combined with smaller numbers of high-quality redshifts taken from deeper spectroscopic datasets. Their accuracy is characterized by to i and for i galaxies, with corresponding catastrophic outlier rates of and (Moutard et al., 2016b, Figure 3). In the case of the faintest galaxies we considered in this paper, namely low-mass quiescent galaxies with at , the photo-z accuracy is better than . Star/galaxy separation (described in Moutard et al., 2016a) discarded 97% of stars while keeping 99% of galaxies.
Galaxy stellar masses were derived as described in Moutard et al. (2016b) with Le Phare using dust-corrected Bruzual & Charlot (2003, hereafter BC03) models of spectral energy distribution (SED), modified to include the effects of emission lines. Rest-frame colours were computed using the nearest observed-frame band in order to minimize dependence on model spectra. The depth of our data allows us to push our analysis to galaxies with around , and around .
2.2 Measurement of the local density
To measure the local density of environment surrounding galaxies in our photo-z sample, we adopted a method similar to Lani et al. (2013) and Malavasi et al. (2016). In brief, we counted the number of galaxies lying in a cylinder of fixed aperture centered on each galaxy for which we measured the density. The cylinder physical depth was set at 1 Gyr, which turned out to be a good compromise to avoid galaxy exclusion and excessive dilution in our case555A depth of 1 Gyr represents a redshift depth of around , i.e., the typical photo-z uncertainty affecting the faintest (quiescent) galaxies of our sample (namely, ) and a redshift depth of at (where ), which then corresponds to the photo-z uncertainty of our faintest galaxies. (for a detailed analysis of the completeness and purity associated with the use of photometric redshifts for reconstructing the galaxy density field, please refer to Malavasi et al., 2016).
It is convenient to define , as the local density normalized by the mean density of the Universe at the same redshift:
where one can see that can also be expressed in terms of the density contrast , as it is commonly defined.666The density contrast is defined by . Quoting the normalized local density measured in cylinder of aperture radius , we can write
where is the number of surrounding galaxies within a cylinder of aperture radius and effective (i.e., non masked) area , and is the total number of galaxies within the corresponding 1 Gyr redshift layer over the entire effective area of the survey (namely, deg in the present analysis777We used the part of the VIPERS-MLS.). We emphasize that, while we made use the normalized local density in the present study, it is generally simply referred as the "local density" in the following, for sake of simplicity.
Aiming to better take advantage of the angular information, we tried several cylinder apertures with radii ranging from 0.3 to 2 physical Mpc (i.e., around the typical galaxy cluster size in the considered redshift range). Figure 1 shows the local density measured in one deg patch of the VIPERS-MLS where we can compare with a map of relaxed galaxy clusters from the XXL survey (Pierre et al., 2016). Namely, we made use of confirmed X-ray clusters from the XXL bright cluster sample (Pacaud et al., 2016). Selected with a flux lower limit of in the keV band of the XMM-Newton satellite, most of XXL bright clusters have masses and redshifts . Unsurprisingly, as one can see, large apertures ( Mpc) tend to smooth the density field, while small apertures ( Mpc) may provide noisier measurements of . We verified that a radius of 0.5 Mpc appears to be a good compromise enabling the detection of over-dense regions whose angular distribution and size match that of bright X-ray clusters, while preventing the measured local density field from being too noisy888This may typically happen when a few galaxies are concentrated over tiny effective areas. by ensuring that over-densities are basically defined from a significant number of galaxies (typically, galaxies for r0.5 Mpc). As a matter of fact, by considering over-dense regions where (blue contours in Fig. 1), we recover 19/19 XXL bright clusters lying in the VIPERS-MLS field at (and 2/2 at ).999Matching was performed according to both the angular position and the redshift, within an angular radius corresponding to XXL cluster radii and by considering the median angular position of our detected over-dense regions, while the redshift tolerance was determined by the typical photo-z uncertainties associated with the median redshift of the over-dense regions.
Figure 2 shows the local density map in the two fields of the VIPERS-MLS at , as measured in 0.5 Mpc radius apertures. The use of photo-z prevents us from being able to trace the subtleties of large-scale structures, like filaments. On the other hand, the method enables the detection of the most massive structures such as clusters (typically seen with , as shown in Fig. 1). As for over-densities having no bright XXL counterparts where the VIPERS-MLS and XXL survey overlap, we cannot exclude some of them to be artefacts. For example, due to the fact that our measure of the local density is projected along the cylinder depth, some large-scale structures may organise along the line of sight (e.g., filament or pair of overlapping groups/clusters).
However, the contribution of such alignments is expected to be low in a large-scale survey101010Even a very large cluster of radius 2 Mpc at would only cover 0.09 deg whilst our survey covers more that 22 deg. and, as shown and discussed in previous studies (see, e.g., Muldrew et al., 2012; Haas et al., 2012; Lani et al., 2013; Malavasi et al., 2016), the uncertainties associated with the use of photometric redshifts tend, on the contrary, to dilute real over-densities along the line of sight,111111In the absence of any significant photometric-redshift bias or catastrophic outlier rate, which is the case in the present study. which therefore makes fake detection of over-dense regions even less probable in our analysis. Moreover, some of these clusters may be not yet virialised clusters (i.e., they are faint or not X-ray emitters) or, even, simply not part of this early XXL release.
We verified that our results were self-consistent by measuring local densities using an alternative approach based on distances to the -nearest neighbour (typically when ). At the same time, densities based on fixed aperture deal naturally well with masked areas (critical in W4) and were shown to correlate very well with high-mass halos (more precisely, when the aperture diameter scales with the virial radius of the halo, typically, Mpc; Muldrew et al., 2012; Haas et al., 2012), which is well suited to our analysis (where the use of photometric redshifts allows the detection of fairly massive galaxy clusters).
3 The NUVrK diagram as a tracer of galaxy evolution
As shown by Arnouts et al. (2013), the rest-frame NUV–r vs. r–K diagram (hereafter NUVrK diagram) is a powerful alternative to the rest-frame UVJ diagram (Williams et al., 2009) to separate quiescent (Q) galaxies from (very dusty) star-forming (SF) galaxies. By extending the wavelength scope of the SED from NUV to NIR, the NUVrK diagram is indeed more sensitive to instantaneous SFR121212UV emission is sensitive to the lifetime of B/A stars, i.e., – Gyr. while being sensitive to stellar ageing and dust attenuation. This results in the enlargement of the so-called green valley, i.e, the space that separates SF and Q galaxies, which allows for a robust selection of star-forming, quiescent, and transitioning (i.e., quenching) galaxies.
3.1 Identifying two quenching channels in the NUVrK diagram
As shown in Moutard et al. (2016a), the volume probed in the VIPERS-MLS is well suited to probe rare populations, which may notably enable us to catch transitioning galaxies that were not observed in smaller surveys. This led us to identify a quenching channel followed by fairly massive galaxies (typically when reaching stellar masses around the characteristic mass ; Moutard et al., 2016b), as recalled in the following.
By quenching channel, we mean a pathway in the rest-frame NUVrK colour diagram that quenching galaxies follow from the star-forming population to the quiescent population. A quenching channel can be associated with an average star-formation history (SFH), which may be highlighted through comparison between colour evolution tracks predicted by stellar-population synthesis models and the actual distribution of galaxy rest-frame colours (see, e.g., Schawinski et al., 2014; Marchesini et al., 2014; Moutard et al., 2016b; Pacifici et al., 2016a). In particular, the NUVrK diagram turns out to be very well suited to distinguish SFHs characterized by different quenching time-scales (Moutard et al., 2016b, Fig. 20) for it is sensitive to very different star lifetimes on each of its axis: Gyr along the rest-frame NUV–r colour (Salim et al., 2005; Martin et al., 2007), hereafter quoted , and Gyrs along the rest-frame r–K colour (Arnouts et al., 2007; Williams et al., 2009), hereafter quoted . Indeed, traces recent star-formation (thanks to rest-frame NUV), while results from the combination of stellar ageing (i.e., the accumulation of generations of low-mass stars, notably traced by rest-frame r) and dust extinction (rest-frame r– being a good tracer of the infrared excess, i.e., the ratio between the UV light absorbed by dust and its re-emission in the infrared; Arnouts et al., 2013): galaxy colours are therefore expected to redden with cosmic time, on average.
In Fig. 3a, we show the NUVrK distribution of our galaxy sample at and the corresponding selection of quiescent and star-forming galaxies, as defined on both sides of the so-called "green valley" where one can identify a line of transitioning (i.e., quenching) galaxies concentrated at that turn out to be fairly massive ( of galaxies with ). We used the upper and lower limits of the time-dependant selection of Q and SF galaxies defined in Moutard et al. (2016b), so that galaxies in transition in the green valley are excluded from our analysis. Namely, Q galaxies were selected with
and SF galaxies with
where is the look-back time at gigen redshift131313E.g., Gyrs at and Gyrs at . (for more detail, please refer to Moutard et al., 2016b, Sect. 5.1).
One can see how a conservative cut at allows us to isolate a population of old quiescent galaxies –i.e., galaxies that exhibit colours typical of evolved (old and dusty) stellar populations– that is expected to be fed by the quenching of fairly high-mass star-forming galaxies reaching (red points in Fig. 3b), while quiescent galaxies exhibiting bluer colours –typical of younger stellar populations– must have followed another quenching channel to turn quiescent (since galaxy colours only redden with time, as explained above). Furthermore, as pointed out in Moutard et al. (2016b), these young quiescent galaxies are essentially low-mass galaxies, and are thus responsible for the upturn observed in the SMF of quiescent galaxies at (magenta triangles in Fig. 3b).
Actually, considering simple e-folding SFH models turns out to be qualitatively and quantitatively well adapted to approximate the average SFHs associated with the flux of quenching galaxies that cross the green valley (see, e.g., Schawinski et al., 2014; Moutard et al., 2016b; Pacifici et al., 2016b), the direction of this flux being supported by the rising fraction of quiescent galaxies that observed with cosmic time from (e.g., Muzzin et al., 2013; Mortlock et al., 2015). Doing so, rejuvenation episodes141414As observed in the local Universe (Salim & Rich, 2010; Thomas et al., 2010) or predicted at higher redshift (Trayford et al., 2016). that may affect the SFR –and so, the colour151515SFR variations only affect , which traces recent star-formation, not .– of individual galaxies are neglected. Indeed, considering an average SFH is equivalent to stacking the SFHs of individual galaxies, which tends to smooth the stochasticity that may characterise their SFRs across cosmic time. As detailed in Moutard et al. (2016b), the average SFH models considered to explain the different quenching channels identified in the NUVrK diagram are characterised by a constant SFR while galaxies are on the star-formation main sequence, until the time of the quenching where the SFR drops as SFR with a quenching time-scales (as represented in panels a and a of Fig. 3).
The quenching channel followed by old high-mass galaxies has thus been shown to be characterised by fairly long time-scales of Gyrs (Figs. 3a, 3a). This corresponds to quenching durations of Gyrs to cross the green valley after leaving the main sequence,161616The set of tracks presented in Moutard et al. (2016b, Fig. 20) is limited to the cases that are able to explain the presence of transitioning high-mass galaxies in the green valley, found to be concentrated at , which is also visible here in Fig. 3a. which is compatible with a strangulation scenario where the cold gas inflows are impeded and the galaxy consumes slowly its reservoir of remaining gas (see, e.g., Peng et al., 2015). In contrast, the quenching channel that would allow us to explain the presence of young low-mass quiescent galaxies must be characterised by shorter time-scales of Gyrs (see Figs. 3a, 3a), which corresponds to quenching durations of only Gyrs to cross the green valley; as a corollary, young low-mass quiescent galaxies are also characterised by a recent quenching.
The NUVrK diagram is therefore a powerful tool to explore the different quenching channels that may be followed by galaxies across cosmic time: fast for (young) low-mass galaxies, slow for (old) high-mass ones.
3.2 Selection of (young) low-mass and (old) massive and ultra-massive galaxies
While young [ < 0.76] quiescent galaxies are essentially low-mass galaxies, their stellar-mass distribution stretches to (Fig. 3b). The relative fraction of these fairly massive young quiescent galaxies is therefore negligible when the stellar mass completeness limit lies below the upturn seen around (namely, when ), which is the case in the VIPERS-MLS at with . Conversely, low-mass  galaxies are mostly young at (Fig. 3c). In other words, on average, young galaxies are low-mass galaxies, and conversely. However, at higher redshift, our stellar mass completeness limit reaches at due to the Malmquist bias. The fraction of fairly massive  galaxies among young galaxies and, conversely, the fraction of old galaxies among low-mass  galaxies then become non negligible. We therefore have to take this into account if we want to focus on low-mass galaxies that are prone to fast quenching at .
Aiming to push our analysis to while ensuring a simultaneous focus on (1) young low-mass galaxies, whose fast quenching is expected to be responsible for the low-mass upturn observed in the SMF of quiescent galaxies, and (2) old high-mass galaxies, whose slow quenching provides the bulk of the quiescent population around , we selected galaxies by combining and stellar mass. As illustrated in Fig. 3d, our refined sample of low-mass () galaxies was therefore selected with
and high-mass () galaxies with
which ensured the low-mass galaxies we considered to be mostly young (i.e., prone to fast quenching), even at (where ).
We also selected a sample of ultra-massive () galaxies with
these galaxies being also essentially old to (Fig. 3c). This later stellar-mass bin with was motivated by the fact that these ultra-massive galaxies seem to be characterised by a peculiar evolution of their number density since with respect to less massive galaxies (see Moutard et al., 2016b, Fig. 15). Being all old galaxies and (almost) all quiescent since this epoch, their evolution is expected to be mostly driven by dry mergers in rich environments. Moreover, these ultra-massive galaxies embody a very advanced stage of galaxy stellar-mass assembly and set therefore a benchmark that is relevant to compare with when considering less massive galaxies.
Combining with the selection of quiescent and star-forming galaxies allowed by the NUVrK diagram (Eqs. 3 and 4), this led to defining five classes of galaxies:
i) quiescent (young) low-mass galaxies, quoted Q, responsible for the upturn observed in the quiescent SMF and expected to have experienced a fast quenching and
ii) star-forming (young) low-mass galaxies, quoted SF, constituting the reservoir of galaxies that might experience such fast quenching;
iii) quiescent (old) high-mass galaxies, quoted Q, responsible for the build-up of the quiescent SMF around and expected to follow a slow quenching channel and
iv) star-forming (old) high-mass galaxies, quoted SF, that are prone to follow this slow quenching channel;
v) and finally ultra-massive galaxies, quoted Q, already old and mostly quiescent (in the redshift range we considered).
4.1 Environments vs. quenching channels
Aiming to explore the impact of environment on the quenching of star formation, we compared the probability distribution functions (PDFs) of the local density measured for the different categories of galaxies defined in Sect. 3.2.
In Fig. 4a, we focus on the quiescent population, divided into low-mass, high-mass and ultra-massive galaxies at . The PDF errorbars were estimated through bootstrap resampling, which accounts for Poissonian uncertainties. In order to characterise each PDF with one single value, that can be seen as the typical local density associated with the corresponding underlying population, we computed the median for each PDF(), denoted (shown with vertical arrows). The first result springing up from the analysis is the confirmation that ultra-massive galaxies clearly reside in the densest environments, with a median local density found to be (grey arrows in Fig. 4a). Unsurprisingly, less massive quiescent galaxies are characterised by lower local densities.
Interestingly, however, low-mass quiescent galaxies clearly appear to be located in denser environments than high-mass ones, especially at where (magenta) and (red). We verified that the distributions were not drown from an identical underlying population (with respect to the projected local density) with the popular and widely used K-S test and the more robust Anderson-Darling test (Anderson & Darling, 1952, 1954).171717The Anderson-Darling test is notably more sensitive to the tails of the distributions (for more details on statistical tests for astronomy and comparisons see, e.g., Babu & Feigelson, 2006; Hou et al., 2009). Both tests confirmed that the two distributions are different at more than 99.99% confidence.
Next, aiming to see if and how the quiescence of galaxies correlates with environment, we compared the PDFs of the local densities measured for quiescent (Q) galaxies with those of star-forming (SF) ones in Fig. 4b. Similarly to Fig. 4a, local density PDFs are shown for low-mass galaxies (light blue) and high-mass ones (dark blue).181818Ultra-massive galaxies are almost all quiescent, only 4/110 galaxies could have been classified as star-forming at (and 7/55 at ), which would prevent us from computing any robust local density PDF. One can first notice a weak –though detectable– trend that tends to make us observe high-mass star-forming galaxies in slightly denser environments than less massive ones. At the same time, star-forming galaxies exhibit local densities that peak near the average density of the Universe, clearly less dense than that of quiescent galaxies regardless of the stellar mass.
As a matter of fact, the difference of median density exhibited by low-mass galaxies according to the star-formation activity is unambiguously the strongest. We have indeed for star-forming galaxies and for quiescent ones at . In fact, we may define the deviation of local density associated with quiescence, , which represents a deviation of in the case of low-mass galaxies (dashed light green arrow in Fig. 4b). As for high-mass galaxies, we also report a non-negligible but weaker local density deviation between star-forming and quiescent galaxies, with and , i.e. (dark green arrow in Fig. 4b).
This confirms that quiescent galaxies are, on average, located in denser environments than star-forming ones, both for low-mass and high-mass galaxies. But the key finding of our analysis is that low-mass galaxies, also identified as prone to being quenched through a fast quenching channel (cf. Sect. 3.2), require a much stronger increase of their typical local density to be observed as quiescent than high-mass galaxies, prone to follow a slow quenching channel, as we discuss later in this paper (Sect. 5.2).
4.2 Local density evolution
Aiming to observe the evolution of typical densities across redshift, we considered the additional redshift bin . The upper redshift limit was set so that it allowed us to probe the excess of low-mass quiescent galaxies at higher redshift while being complete in mass, in addition to ensuring two redshift bins of similar comoving volumes.191919Respectively, Mpc and Mpc at and . To enable comparison of local densities between our redshift bins, we repeated the same analysis than what is presented in Sect. 4.1, but we only considered galaxies more massive than the stellar mass completeness limit of the highest redshift bin, namely , both at and .
In Fig. 5a we show PDF() for the different classes of galaxies we selected (cf. Sect. 3.2), as traced by galaxies with both at (top) and (bottom), where vertical arrows reflect the corresponding values of , similarly to Fig. 4. One may notice how the introduction of a higher stellar-mass completeness limit ( instead of ) reduces the number of low-mass galaxies at (compared to Fig. 4), galaxies with being discarded, and consequently how this affects the measurements of the local density (as traced by the median local density ). The total number of galaxies is indeed reduced when considering instead of and we recall that is normalized by the mean density of the Universe at the same redshift (cf. Sect. 2.2), which proportional to the total number of considered galaxies (Eq. 2). It is therefore expected to measure lower values of when considering a higher stellar-mass limit. The trends we measure are, however, consistent at with or , which confirms our conclusions.
One may thus notice that, as at , ultra-massive galaxies are characterized by the highest local density we measured and that, at lower stellar mass, quiescent galaxies are generally characterised by much higher local local densities than star-forming ones at . In particular, considering quiescent galaxies, Q galaxies may already be characterized by higher local densities than Q ones at , but PDF uncertainties may prevent us from ensuring that low-mass and high-mass quiescent galaxies are characterized by different local densities at .202020At , the confidence given by an Anderson-Darling test of having > is when considering PDF uncertainties (as represented in Fig. 5), but this confidence would drop to in a conservative approach considering PDF uncertainties (i.e., uncertainty containing of the values). Nevertheless, the deviation observed between the local density of quiescent and star-forming galaxies for low-mass galaxies () is already larger than for high-mass galaxies () at . The fact that the quiescence of low-mass galaxies, associated with fast quenching (cf. Sect. 3.2), requires a stronger increase of the local density to be observed than the quiescence of high-mass galaxies is therefore confirmed at as well.
Focussing on the redshift evolution of the typical local density, Fig. 5b shows the evolution of between (median redshift of galaxies in our highest redshift bin, ) and (median redshift of galaxies at ) for the different classes of galaxies we considered. It is thus interesting to notice how constant the typical local density of high-mass star-forming galaxies appears to be constant (), whilst the typical local density of their quiescent counterparts is characterised by a clear increase of between and . This results in the increase of the local density deviation between star-forming and quiescent high-mass galaxies from at to at , therefore essentially due to the fact that the local density of high-mass quiescent galaxies has increased with cosmic time, on average.
At the same time, Fig. 5b reveals an even stronger increase of the typical local density for low-mass quiescent galaxies, with a variation of , which needs to be weighted by the fact that their star-forming counterparts also experienced a small increase of their typical local density with . Still, this results in what appears to be a strong increase of local density deviation observed between star-forming and quiescent for low-mass galaxies from to , namely, from to , compared to high-mass galaxies. While this traces indeed the fact that, on average, the local density of low-mass quiescent galaxies has increased faster with cosmic time than what we observe for high mass galaxies, we will see how the remarkable increases of both (i.e., by definition) and with cosmic time may be explained by the simultaneous modest increase of (see Sect. 5.3).
One may finally notice that might exhibit a very small increase with cosmic time, with between and (from to ). However, one can see how the uncertainties affecting allow for a variation , which prevents us from drawing any conclusion about the local density evolution experienced by ultra-massive galaxies.
We have seen in Sect. 4 how different may be the local density of galaxies depending on whether they are quiescent or not and, above all, depending on the quenching channel they are prone to follow (fast for low-mass galaxies or slow for high-mass galaxies), and then how this may evolve with cosmic time at . In this section, we discuss our results and notably the connection between environment and star-formation quenching that may be highlighted, in particular, the impact of environment on the (fast) quenching of (young) low-mass galaxies and its evolution with cosmic time.
5.1 Ultra-massive galaxies reside in very dense environments
As is obvious in Fig. 5b, ultra-massive galaxies are by far located in the densest environments that were measured in our analysis. These ultra-massive galaxies are almost all quiescent and characterised by old stellar populations from . This makes them good candidates for subsequent growth via (dry) mergers, as already proposed (see, e.g., De Lucia et al., 2006; De Lucia & Blaizot, 2007; Cattaneo et al., 2011; Moutard et al., 2016b; Lee & Yi, 2017; Groenewald et al., 2017).
At the same time, though non-negligible compared to smaller surveys at the similar redshift, the limited number of ultra-massive () galaxies in our analysis (106 at , 48 at ) prevented us from constraining the evolution of their local density at . Constraining such evolution would be of high interest to explore the growth of structures on different scales. Indeed, for instance, a decreasing local density around an ultra-massive galaxy may support a picture where the galaxy merger rate within the host structure is higher than the rate at which new galaxies fall onto the structure, and vice versa.
In any case, the high local densities measured around ultra-massive galaxies support a picture where these galaxies are experiencing a very advanced stage of both galaxy stellar-mass assembly and galaxy clustering.
5.2 The role of environment in the quenching of low-mass galaxies
As described in Sect. 4, when focussing on high-mass galaxies, one can see that quiescent galaxies are characterised by higher typical local densities than star-forming ones, as traced by , both at and (Fig. 4b). This is expected because among high-mass star-forming galaxies, the most massive quench first (see, e.g., Moutard et al., 2016b). At the same time, more massive galaxies are expected to be more clustered on large-scales (typically, what happens around filaments; Malavasi et al., 2017). More massive galaxies are indeed hosted by more massive DM halos, on average, while halo clustering increases with halo mass (given the hierarchical growth of DM structures with cosmic time). In this respect, our study is therefore consistent with many previous studies that have emphasized the fact that quiescent galaxies are preferentially located in denser environments, and in particular concerning massive galaxies with (e.g., Kauffmann et al., 2004; Baldry et al., 2006; Lani et al., 2013; Malavasi et al., 2017; Etherington et al., 2017; Cucciati et al., 2017).
The interest of the present analysis is, however, its ability to disentangle the impact of environment on different categories of galaxies that are prone to follow different quenching channels: slowly quenched (old) high-mass galaxies feeding the quiescent population around , and (young) low-mass galaxies subject to a fast quenching (cf. Fig. 3) responsible for the excess of quiescent galaxies at . Thus, the first remarkable result of our analysis is the fact that these low-mass quiescent galaxies were already located in denser environments than high-mass quiescent galaxies at and probably as of , as observed in the local Universe (e.g., Hogg et al., 2003; Haines et al., 2007).
The role of environment in the quenching of low-mass galaxies is confirmed by the deviation of the typical local density observed between the star-forming and quiescent populations, : besides the fact that quiescent low-mass galaxies appear to be located in much denser environment than their star-forming counterparts, the local density deviation between star-forming and quiescent galaxies is more than twice stronger for low-mass galaxies () than for high-mass ones (). In other words, the quiescence of low-mass galaxies requires a much stronger increase of the local density than the quiescence of high-mass galaxies.
This is therefore consistent with a picture where the upturn observed at low-mass in the SMF of quiescent galaxies is due to the (fast) quenching of (young) low-mass galaxies, due to mechanisms that involve rich environments, as observed in the local Universe (Hogg et al., 2003; Haines et al., 2007; Peng et al., 2012). Our analysis shows that such a picture is also valid at , confirming and complementing the study of Guo et al. (2017) who recently correlated the quenching of low-mass galaxies with rich environments at in the CANDELS fields. While confirming that environment already played a significant role at earlier times, when large-scale structures were less developed, this raises the question of the importance of environmental quenching across cosmic time.
5.3 A rising importance of environmental quenching with cosmic time?
When focussing on low-mass galaxies, we noticed in Sect. 4.2 that the typical local density of SF galaxies slightly increased from to 212121We recall that, to ensure completeness over the entire redshift range , low-mass galaxies have stellar masses . ( in Fig. 5b), which reflects the fact that an increasing fraction of SF galaxies has been characterised by rich environments with decreasing redshift. At the same time, we observed a stronger increase of the typical density for Q galaxies, even already found to be located in much richer environments than their SF counterparts from ( in Fig. 5b). While the increase of the typical local density is due to the growth of large-scale structures that host a growing number of galaxies, the fairly modest increase of observed for SF galaxies is expected if the vast majority of these galaxies are field galaxies, while Q galaxies are preferentially located in rich environments and fully experience the growing number of galaxies within large-scale structures.
Actually, while the number of SF galaxies located in fairly rich environments is small compared to the total number of SF galaxies, it represents a significant number compared to the number of corresponding Q galaxies. The size of the entire SF galaxy population is indeed 10–30 times larger than that of Q galaxies in our sample (cf. Fig 5). For example, the fraction of SF galaxies located in very dense regions where we measure (i.e., 4 times the mean local density of the Universe at the same redshift) is only 1.3% at and 2.7% at , which represents an increase of the number of these galaxies from 195 to 435, whilst the corresponding fraction of Q galaxies increased from 9.5% to 19.6%, but involving fewer galaxies, with 52 and 287 Q galaxies at and , respectively. This highlights how the increasing number of SF galaxies that are characterised by very rich environments is able to feed the strong increase of the number of Q observed in corresponding environments (the contribution of Q to the low-mass population increasing from 21% to 40% between and ). It is, moreover, interesting to note here that SF galaxies remain more numerous than their Q counterparts (79% to 60% at and , respectively) in these very dense regions, as discussed in the next section. In other words, this tends to confirm a picture where the reservoir of low-mass galaxies susceptible to environmental quenching is growing with cosmic time, following the growth of large-scale structures that host a growing number of galaxies.
On the other hand, the fact that the comoving number density of low-mass quenched galaxies has increased with cosmic time does not mean that the corresponding quenching has become more important: the number of low-mass galaxies having quenched via environmental quenching has to be compared with that of high-mass galaxies quenched via mass quenching across cosmic time. In order to quantify the contribution of the environmental quenching channel followed by low-mass galaxies, one may define the low-to-high-mass ratio of the quiescent population at given redshift, , derived as the comoving number density of low-mass environmentally-quenched galaxies relative to that of high-mass mass-quenched galaxies , as
One may thus observe a modest but detectable increase of this ratio from to between and . Yet, this seeming evolution of might be artificial, due to the fact that faint quiescent galaxies are expected to be the firsts to suffer from incompleteness with increasing redshift. As a matter of fact, the completeness limit we adopted (namely, ) ensures our quiescent sample to be more than 95% complete at , but in the particular case of low-mass quiescent galaxies, the completeness can drop to around (against at ). If we assume, in a conservative approach, that all low-mass galaxies suffer from such incompleteness at , the low-to-high-mass ratio of the quiescent population would rather approach at (against at ), which would therefore be consistent with no evolution of with cosmic time at . In other words, the differential incompleteness of low-mass quiescent galaxies at and might be sufficient to explain the increase of that we detected between and .
Nevertheless, the rapid build-up of the low-mass quiescent population observed over the same redshift range from deeper surveys (0.5 dex around , against 0.1 dex around , e.g., in COSMOS; Davidzon et al., 2017) suggests a rising share of low-mass galaxies in the quiescent population, which might plead for a rising importance of the environmental quenching channel (followed by low-mass galaxies) compared to the mass quenching channel (followed by high-mass galaxies). Upcoming large surveys combining deeper optical and near-infrared observations will allow us to verify whether the importance of the environmental quenching channel followed by low-mass galaxies has risen with cosmic time at late epochs.
In any case, our results confirmed that a rising number of low-mass galaxies have been prone to experiencing environmental quenching with cosmic time. The mechanism(s) that may be involved in such environment-driven quenching of low-mass galaxies remain(s), however, a matter of debate, which might be interesting to address in the light of all the elements we gathered so far.
5.4 Composite picture of the environmental quenching channel followed by low-mass galaxies
As discussed extensively in the present paper, the quenching of low-mass galaxies is associated with a strong increase of their local density, which allow us to link the quenching of these galaxies with environmental effects.
At the same time, low-mass quiescent galaxies have been shown to be essentially young quiescent galaxies (i.e., characterised by young stellar populations; cf. Sect. 3.2), which has been shown to require a fast quenching (see, e.g., Schawinski et al., 2014; Moutard et al., 2016b; Pacifici et al., 2016a; Pacifici et al., 2016b). Low-mass quiescent galaxies are therefore recently quenched galaxies. This is consistent with the fact that they exhibit rest-frame colours that are similar to those of post-starburst galaxies (Kriek et al., 2010; Whitaker et al., 2012).222222Our young quiescent population, selected in the NUVrK diagram with rest-frame colours r– < 0.76, overlaps at more than 87% with a sample of the young quiescent galaxies selected in the UVJ diagram with rest-frame colours U–V < 0.9 by Whitaker et al. (2012) as post-starburst galaxies.
It has also been claimed that dwarf satellite galaxies (corresponding to our low-mass galaxies232323We verified that dwarf galaxies of Haines et al. (2007) and our low-mass galaxies overlap at more than 80%.) may be characterised by long quenching time-scales (Haines et al., 2007). That statement was based on the fact that a significant fraction of dwarf satellite galaxies was found to be star-forming in the local Universe, while exhibiting slightly lower star-formation rates than in their field counterparts. Our interpretation is, on the contrary, that those results are consistent with a fast quenching of dwarf satellite galaxies. Indeed, the H equivalent-width distribution measured by Haines et al. (2007, Fig. 5) for dwarf galaxies has only revealed a very small number of transitioning galaxies with respect to that observed in the star-forming and quiescent sequences. And, if dwarf satellite galaxies were slowly quenched, one could expect to statistically observe a significant fraction of them in transition between the star-formation and quiescent sequences, which is not observed.
Rather than slow quenching, those results pleads for a fast quenching of dwarf satellite galaxies in the local Universe, but delayed in onset, since more than 60% of them are star-forming (Haines et al., 2007), which agrees with SMF measurements for central and satellite galaxies in the local Universe where more than 50% of low-mass satellite galaxies are star-forming (Yang et al., 2009; Peng et al., 2012). It is interesting to note that our observations highlight a similar trend at , where 79% and 60% of low-mass galaxies with high local densities () –i.e., prone to fast environmental quenching– are star-forming at and , respectively (cf. Sect. 5.3). Indeed, delayed-then-rapid quenching scenarii, initially proposed in the local Universe to reproduce the SFR distribution of satellite galaxies in clusters (Wetzel et al., 2013; Oman & Hudson, 2016), have recently been shown to be well suited at as well, with an increasing delay before quenching with decreasing stellar mass (Fossati et al., 2017). In such scenarii, the quenching of a satellite galaxy is expected to take a few hundred Myrs, but it occurs several Gyrs after the infall onto the group or cluster. However, as shown by Haines et al. (2007), the fact that dwarf star-forming satellite galaxies exhibit slightly smaller H emission (which traces almost instantaneous SFR) than their field counterparts may highlight the quenching of a part of the star-formation in low-mass galaxies upon or shortly after becoming satellites.
The picture may finally be complemented by the fact that young quiescent galaxies have been shown to be mostly bulge-dominated242424We focussed on galaxies with semi-major axis pixels (i.e, ) at . (Moutard et al., 2016a, Fig. 16), which implies that environmental quenching of low-mass galaxies is probably combined with a rapid morphological transformation. In summary, we may therefore have to consider any scenario supporting a delayed-then-rapid quenching of satellite galaxies, where star formation is suppressed in Gyr (Moutard et al., 2016b) and associated with a simultaneous transformation of galaxy morphology. For example, ram-pressure stripping processes, able to suppress star-formation of a satellite galaxy over 0.2–0.8 Gyrs when it reaches the core of a cluster 2–4 Gyrs after entering it (Mahajan et al., 2011; Wetzel et al., 2013; Muzzin et al., 2014), would require to be associated with tidal stripping harassment to alter the morphology (Moore et al., 1996). Alternatively, the incidence of young low-mass quiescent galaxies in rich environments may be consistent with a major role of mergers within clusters (e.g., Schawinski et al., 2014), by nature compatible with a delayed-then-rapid quenching scenario, while being associated with almost instantaneous transformation of the morphology.
However, it has been shown that the quenching scenario may be quite different depending on the scale of the involved structures (groups or clusters; e.g., Lin et al., 2014). While the aim of the present study was to highlight the role of environment in the fast quenching of low-mass galaxies, the characterisation of the scale at which environmental quenching of low-mass galaxies operates will allow us to specify the physical mechanisms at play.
In an earlier paper (Moutard et al., 2016b), we identified two different quenching channels in the rest-frame NUV–r vs. r–K (i.e., NUVrK) colour diagram: one quenching channel is followed by evolved star-forming galaxies (characterized by old stellar populations) and is expected to be slow, while the other is required to explain the presence of young quiescent galaxies (characterized by young stellar populations) and is expected to be times faster.
The first quenching channel is followed by high-mass galaxies, typically turning quiescent when reaching a characteristic stellar masses of , which is consistent with mass quenching (Ilbert et al., 2010; Peng et al., 2010). In contrast, the other quenching channel is essentially followed by low-mass  galaxies that are responsible for the upturn observed in the SMF of quiescent galaxies, which raised the question of environment role in such quenching channel: is the fast quenching of low-mass galaxies consistent with environmental quenching? Furthermore, the rapid build-up this excess of low-mass quiescent galaxies observed from (e.g., Tomczak et al., 2014; Davidzon et al., 2017) may suggest a rising importance taken by environmental quenching compared to mass quenching with cosmic time (i.e., its rising contribution to the build-up of the quiescent population), as expected in the context of the growth of large-scale structures with cosmic time (e.g., Peng et al., 2010).
In the present paper, we analysed the relation between quenching and environment aiming, in particular, to determine the role played by environment in the quenching of low-mass galaxies. Making use of a galaxy sample complete down to stellar masses of to ( to ) including more than 33,500 (43,000) quiescent galaxies from the VIPERS Multi-Lambda Survey (VIPERS-MLS; Moutard et al., 2016a), we selected galaxies according to the quenching channel they are prone to follow in the NUVrK rest-frame colour diagram while, thanks to accurate photometric redshifts (), galaxy environment was characterized through local density measurements. We summarise our main conclusions below.
In addition to being already mostly quiescent at , ultra-massive  galaxies are characterised by the highest local densities measured in our analysis. This confirms a picture where quiescent ultra-massive galaxies may grow in mass via subsequent (dry) mergers at late epochs (e.g., De Lucia et al., 2006; Cattaneo et al., 2011; Moutard et al., 2016b; Groenewald et al., 2017).
High-mass  quiescent galaxies appear to be generally located in denser environments than their star-forming counterparts. At the same time, the typical local density of high-mass star-forming galaxies appears to be constant between and . This is consistent with a picture where the most massive –and therefore most clustered– among high-mass star-forming galaxies quench first (e.g., Bundy et al., 2006; Ilbert et al., 2010; Davidzon et al., 2013; Moutard et al., 2016b).
Interestingly, we found that low-mass  quiescent galaxies are, on average, characterized by much denser environments than high-mass quiescent galaxies at , and probably already at . Furthermore, the deviation of typical local density observed between quiescent and star-forming low-mass galaxies is always much larger than what can be observed for high-mass galaxies, both at and , which implies that the quiescence of low-mass galaxies requires, on average, a much stronger increase of the local density than for high-mass galaxies. This highlights the lead role of environment in the fast quenching of low-mass galaxies at , consistently with observations made in the local Universe (e.g., Hogg et al., 2003; Haines et al., 2007) and recently at higher redshift (namely, at ; Guo et al., 2017).
In particular, our results confirm that environmental quenching is responsible for the low-mass upturn observed in the SMF of quiescent galaxies at , consistently with what is observed in the local Universe (Yang et al., 2009; Peng et al., 2012).
While the apparent increase of the low-mass galaxy share in the quiescent population that we observed between and may confirm a rising importance taken by environmental quenching over mass quenching with cosmic time, this might be dominated by the differential incompleteness affecting our sample of low-mass quiescent galaxies at and . The simultaneous increase of the typical local density we measured for star-forming low-mass galaxies highlights, however, a clear growth of the reservoir of low-mass galaxies prone to environmental quenching with cosmic time at . Deeper large surveys will soon allow us to confirm whether environmental quenching has become predominant in the feeding of the quiescent population at late epochs, as suggested by the rapid build-up of the SMF low-mass end for quiescent galaxies (Tomczak et al., 2014; Davidzon et al., 2017).
Combining our results with previous studies, we finally refined the composite profile of the quenching process affecting low-mass galaxies. Namely, we have converged to a scenario consistent with the delayed-then-rapid quenching of satellite galaxies (Wetzel et al., 2013), in which low-mass galaxies would remain star-forming after entering the over-dense region to eventually experience a fast quenching in Gyr (Moutard et al., 2016b) while being probably associated with a simultaneous transformation of galaxy morphology (Moutard et al., 2016a). Ram-pressure stripping (Gunn & Gott, 1972), generally put forth, would therefore require to be associated with tidal stripping harassment (Moore et al., 1996) to simultaneously shut star formation down and alter morphology or, alternatively, one may assign the quenching of low-mass galaxies to a major role of mergers within large-scale structures (e.g., Schawinski et al., 2014).
Still, the quenching mechanisms may be quite different depending on the scale of the structures involved in environmental quenching (groups or clusters; e.g., Lin et al., 2014). While our analysis confirmed the role of environment in the fast quenching of low-mass galaxies, the characterisation of the scale at which environmental quenching of low-mass galaxies operates would allow us to specify the physical mechanisms at play.
We gratefully acknowledge the anonymous referee, whose comments helped in improving the clarity of the paper. This research was supported by the ANR Spin(e) project (ANR-13-BS05-0005, http://cosmicorigin.org) and by a Discovery Grant from the Natural Sciences and Engineering Research Council (NSERC) of Canada. This research makes use of the VIPERS-MLS database, operated at CeSAM/LAM, Marseille, France. This work is based in part on observations obtained with WIRCam, a joint project of CFHT, Taiwan, Korea, Canada and France. The CFHT is operated by the National Research Council (NRC) of Canada, the Institut National des Science de l’Univers of the Centre National de la Recherche Scientifique (CNRS) of France, and the University of Hawaii. This work is based in part on observations made with the Galaxy Evolution Explorer (GALEX). GALEX is a NASA Small Explorer, whose mission was developed in cooperation with the Centre National d’Etudes Spatiales (CNES) of France and the Korean Ministry of Science and Technology. GALEX is operated for NASA by the California Institute of Technology under NASA contract NAS5-98034. This work is based in part on data products produced at TERAPIX available at the Canadian Astronomy Data Centre as part of the Canada-France-Hawaii Telescope Legacy Survey, a collaborative project of NRC and CNRS. The TERAPIX team has performed the reduction of all the WIRCAM images and the preparation of the catalogues matched with the T0007 CFHTLS data release.
- Anderson & Darling (1952) Anderson T. W., Darling D. A., 1952, The Annals of Mathematical Statistics, 23, 193
- Anderson & Darling (1954) Anderson T. W., Darling D. A., 1954, Journal of the American Statistical Association, 49, 765
- Arnouts et al. (2002) Arnouts S., et al., 2002, MNRAS, 329, 355
- Arnouts et al. (2007) Arnouts S., et al., 2007, A&A, 476, 137
- Arnouts et al. (2013) Arnouts S., et al., 2013, A&A, 558, A67
- Babu & Feigelson (2006) Babu G. J., Feigelson E. D., 2006, in Gabriel C., Arviset C., Ponz D., Enrique S., eds, Astronomical Society of the Pacific Conference Series Vol. 351, Astronomical Data Analysis Software and Systems XV. p. 127
- Baldry et al. (2006) Baldry I. K., Balogh M. L., Bower R. G., Glazebrook K., Nichol R. C., Bamford S. P., Budavari T., 2006, MNRAS, 373, 469
- Baldry et al. (2012) Baldry I. K., et al., 2012, MNRAS, 421, 621
- Balogh et al. (1997) Balogh M. L., Morris S. L., Yee H. K. C., Carlberg R. G., Ellingson E., 1997, ApJ, 488, L75
- Best et al. (2005) Best P. N., Kauffmann G., Heckman T. M., Brinchmann J., Charlot S., Ivezić Ž., White S. D. M., 2005, MNRAS, 362, 25
- Bruzual & Charlot (2003) Bruzual G., Charlot S., 2003, MNRAS, 344, 1000
- Bundy et al. (2006) Bundy K., et al., 2006, ApJ, 651, 120
- Cattaneo et al. (2006) Cattaneo A., Dekel A., Devriendt J., Guiderdoni B., Blaizot J., 2006, MNRAS, 370, 1651
- Cattaneo et al. (2011) Cattaneo A., Mamon G. A., Warnick K., Knebe A., 2011, A&A, 533, A5
- Chabrier (2003) Chabrier G., 2003, PASP, 115, 763
- Coupon et al. (2015) Coupon J., et al., 2015, MNRAS, 449, 1352
- Croton et al. (2006) Croton D. J., et al., 2006, MNRAS, 365, 11
- Cucciati et al. (2006) Cucciati O., et al., 2006, A&A, 458, 39
- Cucciati et al. (2010) Cucciati O., et al., 2010, A&A, 524, A2
- Cucciati et al. (2017) Cucciati O., et al., 2017, A&A, 602, A15
- Davidzon et al. (2013) Davidzon I., et al., 2013, A&A, 558, A23
- Davidzon et al. (2017) Davidzon I., et al., 2017, A&A, 605, A70
- De Lucia & Blaizot (2007) De Lucia G., Blaizot J., 2007, MNRAS, 375, 2
- De Lucia et al. (2006) De Lucia G., Springel V., White S. D. M., Croton D., Kauffmann G., 2006, MNRAS, 366, 499
- Dekel & Birnboim (2006) Dekel A., Birnboim Y., 2006, MNRAS, 368, 2
- Dekel et al. (2009) Dekel A., et al., 2009, Nature, 457, 451
- Drory et al. (2009) Drory N., et al., 2009, ApJ, 707, 1595
- Etherington et al. (2017) Etherington J., et al., 2017, MNRAS, 466, 228
- Faber et al. (2007) Faber S. M., et al., 2007, ApJ, 665, 265
- Farouki & Shapiro (1981) Farouki R., Shapiro S. L., 1981, ApJ, 243, 32
- Fossati et al. (2017) Fossati M., et al., 2017, ApJ, 835, 153
- Groenewald et al. (2017) Groenewald D. N., Skelton R. E., Gilbank D. G., Loubser S. I., 2017, MNRAS, 467, 4101
- Gunn & Gott (1972) Gunn J. E., Gott III J. R., 1972, ApJ, 176, 1
- Guo et al. (2017) Guo Y., et al., 2017, ApJ, 841, L22
- Guzzo et al. (2014) Guzzo L., et al., 2014, A&A, 566, A108
- Haas et al. (2012) Haas M. R., Schaye J., Jeeson-Daniel A., 2012, MNRAS, 419, 2133
- Haines et al. (2007) Haines C. P., Gargiulo A., La Barbera F., Mercurio A., Merluzzi P., Busarello G., 2007, MNRAS, 381, 7
- Hogg et al. (2003) Hogg D. W., et al., 2003, ApJ, 585, L5
- Hou et al. (2009) Hou A., Parker L. C., Harris W. E., Wilman D. J., 2009, ApJ, 702, 1199
- Hudelot et al. (2012) Hudelot P., et al., 2012, VizieR Online Data Catalog, 2317, 0
- Ilbert et al. (2006) Ilbert O., et al., 2006, A&A, 457, 841
- Ilbert et al. (2010) Ilbert O., et al., 2010, ApJ, 709, 644
- Ilbert et al. (2013) Ilbert O., et al., 2013, A&A, 556, A55
- Ilbert et al. (2015) Ilbert O., et al., 2015, A&A, 579, A2
- Karouzos et al. (2014) Karouzos M., et al., 2014, ApJ, 784, 137
- Kauffmann et al. (2004) Kauffmann G., White S. D. M., Heckman T. M., Ménard B., Brinchmann J., Charlot S., Tremonti C., Brinkmann J., 2004, MNRAS, 353, 713
- Kereš et al. (2005) Kereš D., Katz N., Weinberg D. H., Davé R., 2005, MNRAS, 363, 2
- Kriek et al. (2010) Kriek M., et al., 2010, ApJ, 722, L64
- Laigle et al. (2018) Laigle C., et al., 2018, MNRAS, 474, 5437
- Lani et al. (2013) Lani C., et al., 2013, MNRAS, 435, 207
- Larson et al. (1980) Larson R. B., Tinsley B. M., Caldwell C. N., 1980, ApJ, 237, 692
- Lee & Yi (2017) Lee J., Yi S. K., 2017, ApJ, 836, 161
- Lewis et al. (2002) Lewis I., et al., 2002, MNRAS, 334, 673
- Lin et al. (2014) Lin L., et al., 2014, ApJ, 782, 33
- Mahajan et al. (2011) Mahajan S., Mamon G. A., Raychaudhury S., 2011, MNRAS, 416, 2882
- Malavasi et al. (2016) Malavasi N., Pozzetti L., Cucciati O., Bardelli S., Cimatti A., 2016, A&A, 585, A116
- Malavasi et al. (2017) Malavasi N., et al., 2017, MNRAS, 465, 3817
- Marchesini et al. (2014) Marchesini D., et al., 2014, ApJ, 794, 65
- Martin & GALEX Team (2005) Martin C., GALEX Team 2005, in Colless M., Staveley-Smith L., Stathakis R. A., eds, IAU Symposium Vol. 216, Maps of the Cosmos. p. 221
- Martin et al. (2007) Martin D. C., et al., 2007, The Astrophysical Journal Supplement Series, 173, 342
- Moore et al. (1996) Moore B., Katz N., Lake G., Dressler A., Oemler A., 1996, Nature, 379, 613
- Mortlock et al. (2015) Mortlock A., et al., 2015, MNRAS, 447, 2
- Moustakas et al. (2013) Moustakas J., et al., 2013, ApJ, 767, 50
- Moutard et al. (2016a) Moutard T., et al., 2016a, A&A, 590, A102
- Moutard et al. (2016b) Moutard T., et al., 2016b, A&A, 590, A103
- Muldrew et al. (2012) Muldrew S. I., et al., 2012, MNRAS, 419, 2670
- Muzzin et al. (2012) Muzzin A., et al., 2012, ApJ, 746, 188
- Muzzin et al. (2013) Muzzin A., et al., 2013, ApJ, 777, 18
- Muzzin et al. (2014) Muzzin A., et al., 2014, ApJ, 796, 65
- Oke (1974) Oke J., 1974, ApJS, 27, 21
- Oman & Hudson (2016) Oman K. A., Hudson M. J., 2016, MNRAS, 463, 3083
- Pacaud et al. (2016) Pacaud F., et al., 2016, A&A, 592, A2
- Pacifici et al. (2016a) Pacifici C., Oh S., Oh K., Lee J., Yi S. K., 2016a, ApJ, 824, 45
- Pacifici et al. (2016b) Pacifici C., et al., 2016b, ApJ, 832, 79
- Pandya et al. (2017) Pandya V., et al., 2017, MNRAS, 472, 2054
- Peng et al. (2010) Peng Y.-j., et al., 2010, ApJ, 721, 193
- Peng et al. (2012) Peng Y.-j., Lilly S. J., Renzini A., Carollo M., 2012, ApJ, 757, 4
- Peng et al. (2015) Peng Y., Maiolino R., Cochrane R., 2015, Nature, 521, 192
- Pierre et al. (2016) Pierre M., et al., 2016, A&A, 592, A1
- Salim & Rich (2010) Salim S., Rich R. M., 2010, ApJ, 714, L290
- Salim et al. (2005) Salim S., et al., 2005, ApJ, 619, L39
- Sancisi et al. (2008) Sancisi R., Fraternali F., Oosterloo T., van der Hulst T., 2008, A&ARv, 15, 189
- Schawinski et al. (2014) Schawinski K., et al., 2014, MNRAS, 440, 889
- Scodeggio et al. (2018) Scodeggio M., et al., 2018, A&A, 609, A84
- Scoville et al. (2013) Scoville N., et al., 2013, ApJS, 206, 3
- Thomas et al. (2010) Thomas D., Maraston C., Schawinski K., Sarzi M., Silk J., 2010, MNRAS, 404, 1775
- Tomczak et al. (2014) Tomczak A. R., et al., 2014, ApJ, 783, 85
- Trayford et al. (2016) Trayford J. W., Theuns T., Bower R. G., Crain R. A., Lagos C. d. P., Schaller M., Schaye J., 2016, MNRAS, 460, 3925
- Wetzel et al. (2013) Wetzel A. R., Tinker J. L., Conroy C., van den Bosch F. C., 2013, MNRAS, 432, 336
- Whitaker et al. (2012) Whitaker K. E., Kriek M., van Dokkum P. G., Bezanson R., Brammer G., Franx M., Labbé I., 2012, ApJ, 745, 179
- Williams et al. (2009) Williams R. J., Quadri R. F., Franx M., van Dokkum P., Labbé I., 2009, ApJ, 691, 1879
- Yang et al. (2009) Yang X., Mo H. J., van den Bosch F. C., 2009, ApJ, 695, 900